Ustekinumab for treating active psoriatic arthritis

Clinical effectiveness

4.5 The Committee reviewed the overall clinical effectiveness of ustekinumab. It noted that the evidence for the clinical effectiveness of ustekinumab had been taken from 2 randomised placebo‑controlled trials (PSUMMIT 1 and 2), and acknowledged the need for head‑to‑head studies between ustekinumab and TNF‑alpha inhibitors for psoriatic arthritis. The Committee considered that the evidence suggested that ustekinumab is more effective than placebo after 24 weeks of treatment across a number of joint, skin and soft tissue outcomes. It considered that, although the effect is likely to persist for up to 1 year, there is some uncertainty about this because in the trials people switched from placebo to ustekinumab at week 24. The Committee heard from the clinical experts that ustekinumab appeared to be effective across a wide range of skin and joint outcomes and also soft tissue conditions associated with psoriatic arthritis. The Committee also noted that the results from the PSUMMIT studies suggested there was no statistically significant difference in the clinical effectiveness of ustekinumab compared with placebo between TNF‑alpha inhibitor‑naive and TNF‑alpha inhibitor‑exposed populations for the Psoriatic Arthritis Response Criteria (PsARC) response. The Committee concluded that ustekinumab is clinically effective compared with conventional management, in both TNF‑alpha inhibitor‑naive and TNF‑alpha inhibitor‑exposed populations, but acknowledged that there remains some uncertainty about the long‑term effects of ustekinumab.

4.6 The Committee considered in detail the evidence on the effect of ustekinumab on radiographic outcomes at 24 weeks and 52 weeks. It noted that the effect of ustekinumab compared with placebo appeared to be different to what has been previously observed in clinical trials of golimumab compared with placebo. In particular, ustekinumab appeared to slow the increase (progression) in radiographic score compared with placebo, whereas golimumab (in the NICE technology appraisal guidance on golimumab for the treatment of psoriatic arthritis) had previously been shown to reduce radiographic score from baseline. Furthermore, ustekinumab had not shown a statistically significant difference from placebo in the PSUMMIT 2 study (which included a TNF‑alpha inhibitor‑exposed population). The Committee heard from the company that interpretation of these findings was subject to 4 key difficulties:

4.7 The Committee considered the clinical effectiveness of ustekinumab compared with TNF‑alpha inhibitors in the TNF‑alpha inhibitor‑naive population. The Committee reviewed the findings of the company's mixed treatment comparison (see section 3.7), and noted that the analysis explored the 3 outcomes used as clinical effectiveness inputs in the economic model (Psoriasis Area and Severity Index [PASI] 75, PASI 90 and PsARC response rates). It discussed this analysis with the clinical experts, and was aware of the limitations of the mixed treatment comparison. The Committee concluded that ustekinumab appeared to be less effective than TNF‑alpha inhibitors for PASI 75, PASI 90 and PsARC response, particularly for the joint outcome.

4.8 The Committee also considered the clinical effectiveness of ustekinumab compared with TNF‑alpha inhibitors in the TNF‑alpha inhibitor‑exposed population. It was aware that there was limited clinical trial evidence in this setting. It understood from comments received during consultation that there is some evidence for the effectiveness of TNF‑alpha inhibitors in the TNF‑alpha inhibitor‑exposed population, but was aware that there was not enough evidence to compare ustekinumab and TNF‑alpha inhibitors. The Committee therefore considered the effectiveness of ustekinumab and TNF‑alpha inhibitors compared with conventional management. Although in the PSUMMIT trials there was no difference in clinical effectiveness between TNF‑alpha inhibitor‑naive and TNF‑alpha inhibitor‑exposed populations in terms of PsARC response, the Committee heard from the clinical experts that evidence presented at a conference suggested that the effectiveness of ustekinumab measured using the American College of Rheumatology (ACR) criteria may decrease with increasing numbers of prior TNF‑alpha inhibitors. The clinical experts noted that the diminishing effectiveness of ustekinumab in TNF‑alpha inhibitor‑exposed populations is broadly consistent with clinical experience with the TNF‑alpha inhibitors, which appear to show diminishing effectiveness as the number of prior therapies increases. The Committee heard from the clinical experts that there is some uncertainty about the size of the diminishing effect. The Committee heard estimates for the response rate with second‑line TNF‑alpha inhibitors ranging from 20% to 70%. Conversely, the Committee noted comments received during consultation from a company that manufactures a comparator drug (including evidence from a randomised controlled trial of certolizumab pegol and open‑label and observation studies of adalimumab) that suggested that the lower estimates in this range may be too low. The Committee also considered whether there may be any variation in clinical effectiveness depending on the reason for withdrawal of the first TNF‑alpha inhibitor (for example, initial lack of efficacy, gradual loss of efficacy over time or adverse reactions), but it acknowledged that there was not enough evidence for this aspect to be considered further. The Committee concluded that there is still uncertainty about the relative effectiveness of ustekinumab and TNF‑alpha inhibitors in people who have previously had TNF‑alpha inhibitors.

4.9 The Committee queried whether both the 45‑mg and 90‑mg doses of ustekinumab might potentially be used in clinical practice and, if so, how the doses might be used. It noted that the marketing authorisation for ustekinumab in psoriatic arthritis indicates that 45 mg may be used for all patients and 90 mg may be considered in people who weigh more than 100 kg, concluding that this permits, but does not require, a weight‑based dosing strategy. The Committee also noted that it had not been shown detailed evidence on the relative effectiveness of the 2 doses in people of different weights. The Committee considered the evidence in the European public assessment report published by the European Medicines Agency (EMA), which noted that systemic exposure to ustekinumab (that is, the concentration of ustekinumab in the serum) is similar in people who weigh more than 100 kg and have ustekinumab 90 mg, compared with people who weigh less than 100 kg and have ustekinumab 45 mg. Moreover, the EMA noted that the efficacy of ustekinumab 90 mg (in terms of ACR 20 response) was higher than ustekinumab 45 mg, particularly in people who weigh more than 100 kg, in the PSUMMIT 1 study, although not in PSUMMIT 2. The Committee heard from the company that the dose–response effect based on weight for psoriatic arthritis may not be as strong as seen in psoriasis and that the differences between doses were not statistically significant. The Committee also considered evidence from the Evidence Review Group (ERG), which suggested that there was no statistically significant difference in clinical effectiveness between the higher and lower doses, although it was noted that this did not imply the doses are equivalent. The Committee heard from the clinical experts that if ustekinumab were recommended, they would anticipate using both the 45‑mg and 90‑mg doses in clinical practice (rather than only the 45‑mg dose). The Committee acknowledged that there is no clear evidence to support the use of a strict weight‑based dosing strategy, although it concluded that both the 45‑mg and 90‑mg doses would be expected to be used in clinical practice.

Cost effectiveness

4.10 The Committee considered the structure, assumptions and results in the company's economic model and the critique presented by the ERG. In particular, it discussed key assumptions about the improvement, rebound and progression of joint symptoms, the effect of conventional management on skin symptoms, the use of the utility equation, the timing of the assessment of response and the sequencing of biological treatments in the TNF‑alpha inhibitor‑exposed population. It then reviewed the effect of these assumptions on the cost‑effectiveness estimates for ustekinumab. The Committee also considered the additional analyses incorporating the patient access scheme, presented during the rapid review.

4.11 The Committee noted that the assumptions about the improvement, rebound and progression of joint symptoms (as captured using the Health Assessment Questionnaire Disability Index [HAQ‑DI]) were key drivers of the economic model. It noted that the approach used in the company's model (in which HAQ‑DI improved by a fixed amount, giving an improved HAQ‑DI score that was maintained at a constant level for the duration of biological treatment, rebounded after treatment withdrawal and then gradually deteriorated during conventional management [see section 3.11]) was consistent with the models used in previous NICE technology appraisal guidance on TNF‑alpha inhibitors for psoriatic arthritis (etanercept, infliximab and adalimumab for the treatment of psoriatic arthritis and golimumab for the treatment of psoriatic arthritis). It heard from the clinical experts that HAQ‑DI is an acceptable and sensitive treatment outcome measure. The clinical experts noted that the HAQ‑DI rebound effect on withdrawal of TNF‑alpha inhibitors is not necessarily immediate, but may be associated with a lag of approximately 6 to 12 months, which may also apply to ustekinumab. Furthermore, the Committee considered it possible that the assumption that people have a fixed improvement in HAQ‑DI that is maintained during treatment may not apply to ustekinumab, because it has a different mechanism of action to TNF‑alpha inhibitors. The observed differences in radiographic progression (see section 4.6) may provide some support for this suggestion. Conversely, the Committee understood that the radiographic progression results must be interpreted with caution, and also noted evidence from the PSUMMIT 1 study on HAQ‑DI scores with ustekinumab after 52–100 weeks that did not suggest a substantial worsening over time (see section 3.28; long‑term analyses of HAQ‑DI scores in PSUMMIT 2 were not available at the time NICE technology appraisal guidance 313 was prepared). The Committee considered it possible that there may be some worsening of HAQ‑DI score during ustekinumab treatment, and that this would be likely to decrease the cost effectiveness of ustekinumab, although the size of this effect is unknown. The Committee acknowledged that there is a lack of robust evidence to reliably inform these assumptions, but would have liked to have seen an assessment of the effect on the model results of worsening HAQ‑DI over time during ustekinumab treatment. The Committee concluded that uncertainty remains as to how well the HAQ‑DI assumptions apply to ustekinumab, but considered that the assumptions in the model were a sufficient basis on which to make a decision.

4.12 The Committee considered the way in which the effect of conventional management on skin symptoms had been modelled. It noted that the company's original, post‑clarification and post‑consultation models (see section 3.12) assumed that conventional management strategies did not affect skin symptoms, but heard that the ERG's clinical adviser stated that in practice, DMARDs such as methotrexate often improve psoriasis symptoms. During consultation the Committee received additional information, from a company that manufactures a comparator drug, on the effect of conventional management on skin symptoms, taken from a study of adalimumab. In the rapid review, the Committee noted that the company updated its model to incorporate the effect of conventional management on skin symptoms. It heard from the ERG that the modelling approach was consistent with the approach taken for biological treatments in the original model, and understood that the ERG considered this mostly reasonable. The Committee concluded that it was appropriate to include the effect of conventional management on skin symptoms in the economic model.

4.13 The Committee noted that the company's base‑case analysis was based on utility scores derived using a previously published equation used in the NICE technology appraisal guidance on etanercept, infliximab and adalimumab for the treatment of psoriatic arthritis and golimumab for the treatment of psoriatic arthritis. However, it also noted that health‑related quality‑of‑life evidence had been captured directly in the PSUMMIT studies through the 36‑item Short‑Form Health Survey (SF‑36). The Committee understood that the company used the SF‑36 data to derive an alternative utility equation, and that the impact of this approach on the model results was examined in a sensitivity analysis. However, it further noted that this alternative utility equation was subject to uncertainty, because of a need to map from SF‑36 to EQ‑5D using evidence from people without psoriatic arthritis. The Committee considered that all health‑related quality‑of‑life evidence from the clinical trials – including the SF‑36 data – should ideally be used if possible. However, because of the uncertainty in the newer utility equation, and the fact that the effectiveness of ustekinumab in the PSUMMIT trials was captured through the HAQ‑DI and PASI scores, the Committee concluded that using the previously published equation would be more appropriate and would support a consistent approach between appraisals.

4.14 The Committee considered the appropriateness of assessing treatment responses at week 12 for TNF‑alpha inhibitors and conventional management, and week 24 for ustekinumab. It heard from the clinical experts that there is some uncertainty about when ustekinumab begins to take effect, although its onset of action may be slower than TNF‑alpha inhibitors. The clinical experts stated that DMARDs such as methotrexate often show little or no effect after 12 weeks, and if they do provide benefits these may arise with longer treatment. It was suggested by the clinical experts that there is no specific reason why ustekinumab and TNF‑alpha inhibitors should be assessed at the same time point, because they are different treatments, although the use of different time points in the economic model is likely to favour ustekinumab. The Committee heard during consultation that the British Society for Rheumatology guidelines define a therapeutic trial of DMARDs as at least 12 weeks, although it noted that this did not preclude assessment of response at 24 weeks. The Committee concluded that, for pairwise comparisons between ustekinumab and conventional management, the treatment response should ideally have been assessed at the same time point. The Committee had a preference for assessing treatment response at 24 weeks for both ustekinumab (in line with its summary of product characteristics and the primary efficacy analysis of the PSUMMIT 1 and 2 studies) and conventional management. However, it understood that the company considered that there was no intrinsic reason why the timing of the assessment for ustekinumab and conventional management in the economic model must be the same, and that assessing the response to conventional management at 24 weeks would be inconsistent with NICE technology appraisal guidance on etanercept, infliximab and adalimumab for the treatment of psoriatic arthritis and golimumab for the treatment of psoriatic arthritis. The Committee noted that assessing the response to conventional management at week 24 rather than week 12 increased the incremental cost‑effectiveness ratio (ICER) for ustekinumab.

4.15 The Committee considered the impact of sensitivity analyses and scenario analyses on the results of the economic model. It noted that the company and the ERG presented a number of analyses (see sections 3.12, 3.24, 3.30 and 3.40). The Committee noted that the model was highly sensitive to assumptions about the HAQ‑DI score. In addition to the key assumptions explored in sections 4.11–14, the Committee noted that there were further assumptions that were subject to uncertainty, but which had little impact on the results of the model. It noted that the company's weight‑based dosing approach did not appear to substantially influence the results of the economic model. The Committee also noted that the longer time horizon in the company's model compared with previous models did not dramatically affect the results, although it highlighted that a 40‑year time horizon was preferable to ensure consistency with previous appraisals. During consultation, a consultee noted that the withdrawal rate for ustekinumab had been taken from studies of TNF‑alpha inhibitors, and that it may be more appropriate to use the withdrawal rate from the PSUMMIT studies. The Committee heard from the company that the withdrawal rate in PSUMMIT 1 was lower than that included in the model. However, the Committee was also aware that this rate was derived from a study of 2 years' duration and the long‑term withdrawal rate for ustekinumab is unknown, but the economic model had a lifetime time horizon. The effect of this assumption on the model was not presented, but the Committee considered that if the withdrawal rate were lower than 16.5%, the ICERs for ustekinumab might be expected to decrease by a small amount. The Committee concluded that the weight‑based dosing assumption, the time horizon and the withdrawal rate were not key drivers of the economic model and they did not have a substantial effect on the ICERs.

4.16 The Committee considered the appropriateness of appraising ustekinumab 45 mg alone, in light of the additional analyses presented by the company and the ERG (see sections 3.26 and 3.27). Based on input from the clinical experts, the Committee considered that both the 45‑mg and 90‑mg doses were likely to be used in clinical practice (see section 4.9). The Committee also considered whether appraising ustekinumab 45 mg alone could lead to unfair or discriminatory recommendations, if the higher dose were more effective in people weighing more than 100 kg. The Committee concluded that, based on the likely use of ustekinumab in clinical practice and the potential for effectiveness differences between the doses (particularly in people weighing more than 100 kg), it would not be appropriate for it to consider ustekinumab 45 mg alone.

4.17 The Committee considered the analyses incorporating the patient access scheme provided by the company and the ERG for the rapid review of NICE technology appraisal guidance 313. It noted that the ERG had corrected errors in the company's model, and considered these corrections appropriate. The Committee considered that it would have been preferable to include the additional costs associated with the patient access scheme in the model, although it understood that the effects of these costs on the results would be small. The Committee noted that the company had incorporated the effect of conventional management on skin symptoms in its base case, and that the ERG had incorporated this assumption along with the assessment of treatment response at week 24 for both conventional management and ustekinumab and a 40‑year time horizon in its scenario analysis. The Committee considered that the probabilistic ICERs, when available, were more informative than the deterministic ICERs. It concluded that the ERG's scenario analysis (see section 3.40) reflected the Committee's preferred assumptions and therefore provided the most informative results and the most plausible ICERs, although it noted that the ICERs would decrease if the response to conventional management were assessed at week 12.

4.18 The Committee considered the cost effectiveness of ustekinumab in the TNF‑alpha inhibitor‑naive population. It considered the incremental analysis to be appropriate for most people with psoriatic arthritis, for whom TNF‑alpha inhibitors are the most appropriate comparator (see section 4.2). The Committee noted that, with the patient access scheme, ustekinumab was the lowest‑cost biological treatment but was extendedly dominated (that is, was more expensive and less effective than a combination of 2 comparators). Moreover, the Committee noted that the cost‑effectiveness analyses were subject to uncertainty because of the potential effect of a possible increase in HAQ‑DI during ustekinumab treatment, which would be expected to reduce the cost effectiveness of ustekinumab. The Committee concluded that ustekinumab is not a cost‑effective option, compared with TNF‑alpha inhibitors, for treating psoriatic arthritis in people who have not previously had TNF‑alpha inhibitors.

4.19 The Committee also considered the cost effectiveness of ustekinumab in people who have not previously had TNF‑alpha inhibitors and for whom TNF‑alpha inhibitors are contraindicated. It considered that this population comprises people for whom a TNF‑alpha inhibitor would otherwise be considered (as per the criteria described in etanercept, infliximab and adalimumab for the treatment of psoriatic arthritis and golimumab for the treatment of psoriatic arthritis, see section 4.1). In this population, the Committee considered that conventional management is the most appropriate comparator. It emphasised that ustekinumab is innovative for this population, because it potentially fulfils an important unmet need. It noted that the number of people in this situation was unknown and may be very small, although it was aware of the need to identify subgroups for which the technology may be cost effective. Moreover, the Committee noted that no evidence had been presented specifically for this population; the available evidence was drawn from the PSUMMIT studies, which were likely to have included a mixture of people for whom TNF‑alpha inhibitors would and would not be appropriate. The Committee noted that when the patient access scheme was applied and its preferred assumptions were incorporated, the most plausible ICER was £21,900 per quality‑adjusted life year (QALY) gained, compared with conventional management. Although it considered that this ICER was still subject to uncertainty because of the possible increase in HAQ‑DI during ustekinumab treatment (see section 4.11), the Committee noted that the ICER would be lower if the response to conventional management were assessed at week 12. It was also conscious that there is considerable unmet need in this population and that ustekinumab is an innovative treatment in this setting. The Committee therefore concluded that ustekinumab is a cost‑effective option for treating psoriatic arthritis in people who have not previously had TNF‑alpha inhibitors and for whom TNF‑alpha inhibitors are contraindicated.

4.20 The Committee considered the cost effectiveness of ustekinumab in the TNF‑alpha inhibitor‑exposed population. It noted that for most people, an alternative TNF‑alpha inhibitor is the most appropriate comparator, which was not presented in the company's original submission. The Committee therefore considered the exploratory sequencing analysis initially presented by the ERG, noting that the analysis was reproduced by the company in its rapid review model. This analysis assessed the cost effectiveness of ustekinumab and TNF‑alpha inhibitors when used as second‑line treatments, when first‑line treatment with a TNF‑alpha inhibitor had failed because of lack of efficacy or adverse reactions, and comprised an incremental comparison of ustekinumab, TNF‑alpha inhibitors and conventional management. The ERG and the Committee acknowledged that this analysis is subject to considerable uncertainty. This was because there was no distinction between people whose disease showed no initial response to TNF‑alpha inhibitors and those for whom TNF‑alpha inhibitors failed during long‑term treatment; the clinical experts noted that these groups represent 2 distinct populations. Nevertheless, the Committee considered that this exploratory analysis provided useful information for establishing a full picture of the cost effectiveness of ustekinumab. The Committee considered that the most informative scenario was the one in which the first‑line TNF‑alpha inhibitor failed because of lack of efficacy and clinical effectiveness data for ustekinumab were taken directly from PSUMMIT 2, and noted that this analysis was the most favourable for ustekinumab. With the patient access scheme and the preferred assumptions incorporated, the Committee noted that in the incremental analysis, the most plausible ICER for ustekinumab was £25,400 per QALY gained (compared with conventional management). The Committee was aware that the ICER was subject to uncertainty because of the possible increase in HAQ‑DI during ustekinumab treatment, and noted that the ICER would decrease if the response to conventional management were assessed at week 12. The Committee understood that this analysis was uncertain, but concluded that it was reasonable to recommend ustekinumab as a treatment option for people who have previously had TNF‑alpha inhibitors and for whom treatment with a subsequent TNF‑alpha inhibitor is appropriate.

4.21 The Committee considered the cost effectiveness of ustekinumab in the TNF‑alpha inhibitor‑exposed population, looking specifically at people for whom TNF‑alpha inhibitors as a class had failed. It understood the important unmet need for people in this situation. The Committee also understood that there is limited evidence for this population, because the PSUMMIT 2 study included a mixture of people for whom subsequent TNF‑alpha inhibitors would and would not be appropriate. It highlighted that conventional management is an appropriate comparator in this population. With the patient access scheme and preferred assumptions incorporated, the Committee considered that the most plausible ICER compared with conventional management in the TNF‑alpha inhibitor‑exposed population was £25,300 per QALY gained. Similarly to the TNF‑alpha inhibitor‑naive population, the Committee understood that this ICER was still subject to uncertainty because of the possible increase in HAQ‑DI during ustekinumab treatment, although it noted that the ICER would decrease if the response to conventional management were assessed at week 12. The Committee was also aware of the considerable unmet need in this population. The Committee concluded that ustekinumab is a cost‑effective option for treating psoriatic arthritis in people who have previously had TNF‑alpha inhibitor therapy and for whom TNF‑alpha inhibitors as a class have failed.

4.22 The Committee discussed the recommendation to stop treatment based on an inadequate PsARC response in the NICE technology appraisal guidance on etanercept, infliximab and adalimumab for the treatment of psoriatic arthritis and golimumab for the treatment of psoriatic arthritis. The Committee noted that the economic analyses (in all populations) were based on the assumption that people whose psoriatic arthritis has not shown an adequate PsARC response at 24 weeks stop treatment with ustekinumab. The Committee considered that the recommendation to stop treatment based on an inadequate PsARC response (as defined in the NICE technology appraisal guidance on etanercept, infliximab and adalimumab for the treatment of psoriatic arthritis and golimumab for the treatment of psoriatic arthritis) was also appropriate for ustekinumab (assessed at 24 weeks). It noted that some people may have physical, sensory or learning disabilities or communication difficulties that could affect their responses to components of the PsARC, and concluded that this should be taken into account when using the PsARC.

4.23 The Committee considered evidence from the company on the innovative nature of ustekinumab. It heard from the clinical experts that they considered ustekinumab to be an innovative technology, because it is in a different class to the TNF‑alpha inhibitors and targets a different inflammatory pathway. They considered ustekinumab to be a particularly valuable treatment option in people for whom TNF‑alpha inhibitors are not appropriate. The Committee noted that there is an important unmet need in people for whom TNF‑alpha inhibitors are inappropriate or not effective. However, the Committee considered that, although the introduction of TNF‑alpha inhibitors represented a 'step change' in managing psoriatic arthritis, evidence from the mixed treatment comparison suggested that ustekinumab may be less effective than TNF‑alpha inhibitors, and so ustekinumab does not represent a clear‑cut further step change compared with TNF‑alpha inhibitors. The Committee also considered the innovative nature of ustekinumab for people for whom TNF‑alpha inhibitors are inappropriate. It understood that some of the contraindications for TNF‑alpha inhibitors also apply to ustekinumab, so ustekinumab would not be appropriate for all people for whom TNF‑alpha inhibitors are unsuitable. The Committee considered that all of the health‑related benefits associated with ustekinumab had been adequately captured in the economic model, and no changes to the recommendations were needed for that reason.

4.24 The patient expert highlighted that people with psoriatic arthritis often have concerns about the long‑term safety of treatments for this condition. The Committee was aware of registers that collect evidence on the long‑term treatment outcomes with TNF‑alpha inhibitors for rheumatoid arthritis and psoriasis. The patient expert and the clinical experts emphasised the importance of collecting long‑term data on psoriatic arthritis specifically. The Committee concluded that long‑term evidence on the effectiveness and safety of biological treatments for psoriatic arthritis would be valuable.

Summary of Appraisal Committee's key conclusions